De retibus nummorum. Interconnectivity in the late-Celtic world: the example of coinage

• Applicants: Dr. Caroline von Nicolai; Dr. Karsten Tolle
• Subject Area: Prehistory and World Archaeology ; Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 534092861

The interdisciplinary research project de retibus nummorum (de re num) aims to reconstruct the emergence and development of the Celtic monetary economy, communication networks and economic areas. The basis for this are Celtic coins of the 3rd to 1st centuries B.C., which will be studied by means of typological distribution and die links, as well as by analysing the coins’ find contexts. This disparate information will be incorporated into a social network analysis, a method currently used in the social sciences, but not yet established in numismatics. Social network analysis uses quantitative analysis to explore the connections and dependencies of individual actors - here sites and coins - in a social system. In our example, the sites, in particular settlements, form the nodes of the network, whereas matching types or dies represent the edges. Coins from secured archaeological find contexts are therefore most relevant for the study. The results of the social network analysis will be presented in a Geographic Information System to capture the spatial dimension of coin circulation in selected time slices. In order to leverage as complete a data basis as possible, the project uses computer-assisted approaches based on Artificial Intelligence with the aim of semi-automatically assigning a large number of coin image datasets to the respective coin types or dies. Celtic coins, in particular coins of often varying degrees of preservation from finds, pose a particular challenge for Artificial Intelligence due to their diversity and unbalanced datasets. Already in a preliminary project (ClaReNet), funded by the BMBF, we were able to show exemplarily that these approaches (based on Deep Learning as well as Feature Detection algorithms) have very good potential for accelerating the traditionally very time-consuming, manual process of identifying both typological correspondence and die links, despite the diversity of the coins. The goal of the project is to further refine this approach in close collaboration between numismatics and computer science, and to create open-source tools that can be further used by others accordingly, for example for semi-automated recognition of other categories of archaeological finds. The coin photographs produced by the project will be made available, annotated with the relevant metadata, to a broad public in the virtual union catalog Online Celtic Coinage.

• DFG Programme: Research Grants
• Co-Investigator: Dr. David Wigg-Wolf